IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 16, NO.

1, MARCH 2001 81

Inspection of Stator Cores in Large Machines with

a Low Yoke Induction Method—Measurement
and Analysis of Interlamination Short-Circuits
Zlatimir Posedel

Abstract—Interlamination short-circuits can cause major

damage to electrical machines. Especially endangered are large
turbomachines with high yoke width and correspondingly high
interlamination voltages. These generate during operation, at
certain interlamination short-circuit contact resistance’s, high
short-circuit currents and lead in the worst case to “core
melting.” Generally the stator core for interlamination short-
circuits is inspected under application of the high induction
method, which often cannot give a satisfactory report on the
lamination insula- tion condition. In particular the large Fig. 1. Stator core magnetizing with low induction.
magnetizing expense of the stator core (large voltage and current
values of the magnetizing cable, availability of a strong current
source) has proved to be disadvantageous. Furthermore, this correspondingly damage the machine severely. The measure-
method permits only a localization of hot spots on the surface of ment of the core tooth leak field enables the maximum
the tooth in the case of an assembled stator winding. All other possible contact power at the interlamination short-circuit spot
interlamination short-circuits, especially far more dangerous to be de- termined. If the ascertained contact load exceeds a
ones in the slot wall or on the slot bottom respectively, are not
accessible with this inspection. Also the interlamination short defined crit- ical power; the faulty spot is considered as a
circuit with low contact resistance generated at the contact point dangerous interlam- ination short-circuits. The critical power is
little heat and therewith the low temperature. For these reasons, defined as the lowest limit where “core melting” could arise.
there is a need to develop a safe measuring method, which This critical power was determined theoretically as well as in
enables on one hand all interlamination short-circuits to be
trials. The interlamination isolation defects in the core end
registered, and on the other hand a quantita- tive assessment of
the danger of the interlamination short-circuits for the machine. packets caused from the axial machine field can be also well
For nearly 20 years a measuring method with lower yoke in- detected. With the help of the calibration procedure, the
duction has been used without disadvantages of the high induc- following is possible:
tion method. By this method an interlamination short-circuit is — Recording of the entire core tooth leak field over the stator
de- tected with a measuring coil. The signal has been interpreted
in terms of current. With this interpretation of the measuring signal teeth (“fingerprints”).
it is not possible to take a meaningful consistent quantitative — The recorded data serve for qualitative assessment of the
analysis of the interlamination short-circuit. For a correct analysis design configuration of the stator and for a trend analysis.
we intro- duce a calibration procedure and data processing — Recognition of all interlamination short-circuits by the
algorithm. The method with introducing a calibration procedure
determination of characteristic data (amplitude, phase
permits a com- plete analysis of the lamination insulation, both
quantitatively and qualitatively. The analysis of the measuring angle).
signal, the mechanism of “core melting” and the comparison of — Localization of dangerous interlamination short-circuit
the measuring methods will be treated in detail in this article. spots by the determination of characteristic data (size,
contact power, field gradient and position).
I. INTRODUCTION — Inspection of the interlamination short-circuits at
genera- tors with assembled rotor.
I N THE case of an interlamination short-circuit, the lam-
ination voltage is short-circuited between at least 2 lami-
nations. Through the measurement of the core tooth leak field
— Quick repair of the iron core with continuous checking of
the repair steps.
with introducing the calibration procedure it is possible to de-
tect all interlamination short-circuits. At this interlamination II. STATOR INTERLAMINATION SHORT-CIRCUIT
short-circuit spot, an electrical contact power will arise, which DETECTION WITH LOW INDUCTION METHOD
heats up the defective spot. Under certain contact conditions, A. Magnetizing
the power consumption at the defect spot can lead to a melting The iron core is weakly magnetized with a coil, which is
of this contact. Further spreading can cause “core melting” placed around the stator core (Fig. 1). Because the yoke induc-
and tion in this measurement is about 5% of the rated induction,
the low voltage net is sufficient for feeding the induction
Manuscript received September 30, 1998; revised April 5, coil. A variable transformer from the low voltage socket
1999. The author is with ABB Power Generation Ltd.
Publisher Item Identifier S 0885-8969(01)02660-2.
outlet (220/110 V, 50/60 Hz) performs the ring magnetization.
Through compensation of the inductive reactive current by
0885–8969/01$10.00 © 2001 IEEE
82 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 16, NO. 1, MARCH 2001

Fig. 3. Leak field of the fault current.

— Total magnetic induction in the measuring coil

The constant value of the total magnetic induction does not

depend on the coil dimensions.
The induced measuring voltage can be further written as:
Fig. 2. Tooth leak field detection with measuring coil.

In the measuring coil, in the absence of interlamination short-

means of parallel connected capacitors, the feeding current
circuits, a voltage will be induced which is linearly dependent
can be reduced. The maximum feeding current will not exceed
on the magnetizing current. As the magnetizing current does not
20 A for this measurement.
have a sinusoidal shape (due to the hysteresis), the induced mea-
The magnetization of the stator core in the case of machines
suring voltage will not be a sinusoidal signal. The measured
with assembled rotor, especially in case of large turbomachines,
voltage of the measuring coil, when the iron core is free
is preferably done with a low voltage source, which is connected
of interlamination short-circuits, is proportional to the
between two machine sides to the shaft. The rotor must be
magnetic potential between the coil ends respectively to the
com- pletely insulated against earth on one side of the
magnetizing current. The measuring coil serves as a magnetic
machine. De- fective insulation on one side of the machine
potential mea- suring device (Rogowski coil or Chattock
may be observed through measurement of the shaft current
potentiometer) of the linear integral between the end points of
with a Rogowski coil and with the control of the shaft earth
the measuring coil.
potential.
If there is an interlamination short-circuit, the fault current
will induce a voltage in the measuring coil (Fig. 3).
B. Analysis the Measuring Signal Because of the small interlamination short-circuit length,
In contrast to traditional iron core examination at rated induc- which is much smaller than the coil width, the total magnetic
tion, where hot interlamination short-circuits (“Hot spots”) are induction is de- pendent on the dimension of the coil
registered (infra-red camera, hand checking), this method uses
one or two coils assigned on the surface in order to determine
the magnetic tooth leak field Φ str .The measuring coil with the The measuring coil can not be interpreted as Chattock mag-
length , the cross-section and the number of winding ω is netic potentiometer for the interlamination short current as de-
mounted in the iron core between two teeth and moved along scribed in El Cid tests [1]. The interlamination short-circuit
the slots (Fig. 2). The induced voltage of the measuring coil if cur- rent cannot be determined from the measured voltage.
there are no interlamination short-circuits is: The in- terpretation of the measuring signal is a significant
weakness in the El Cid test. Since a meaningful and consistent
quantitative analysis of the measuring signals is not possible.
The measure- ment signal is dependent on the fault current
magnitude and the length of the interlamination fault. Because
With fulfilled the following conditions: of the small inter- lamination short-circuit length, which is
— Number of windings per length-unit much smaller than the coil width, the interlamination short-
circuit current cannot be determined from the measured voltage.
This is a significant lack of the El Cid tests [1]. The
measurement signal dependent from the product value of the
— Coil cross-section along the measuring coil fault current magnitude and the length of the interlamination
fault. The measure voltage at different in- terlamination short-
circuits can be seen in Fig. 4.
POSEDEL: INSPECTION OF STATOR CORES IN LARGE MACHINES WITH A LOW YOKE INDUCTION 83
METHOD

Fig. 4. Time pattern of the measured voltage at different interlamination

short-circuits: — without interlamination short-circuits; - - - with 5 mm
interlamination short-circuits; - - - with 10 mm interlamination short-
circuits.

Fig. 6. Amplitude and phase angle changes at interlamination short-circuits.

Fig. 5. Vector diagram with unchanged yoke field: u Measured voltage

=

without fault; u Measured voltage with interlamination short-circuits;

=

= Phase angle without interlamination fault; Phase angle with

=

interlamination faults; ' Phase angle of the fault current.

=

u = u u

= :

C. The Recognition of the Interlamination Short-Circuits

An interlamination short-circuit is characterized by an
increased phase angle and amplitude of the measured voltage
(under the assumption that there is no effect of the interlami- Fig. 7. Calibration arrangement.
nation short-circuit current to the magnetization in the yoke)
(Fig. 5).
The phase and amplitude deviations and , caused
by the lamination short-circuit currents, are registered over the
slot length and stored in a computer. The measured phase
angle signal is independent of the yoke induction in the range
of a magnetic yoke induction between 0.003 T and 0.7 T. The Fig. 8. Measured signal change during calibration and interlamination
mea- suring signal is fully reproducible and serves as a short-circuit.
“fingerprint” of the stator core lamination state. The
measurement is used for trend analysis (Fig. 6). signal change depends on the magnitude of the interlamination
Without an interlamination short-circuit, the voltage and short-circuit current and its length :
phase angle of the measuring signal vary because of the
varying magnetic potential in the circumferencial direction.
This varying magnetic potential is characterized by reduced
ampli- tude accompanied by increased phase angle of the The data registered with the measuring probe along the slots
measured voltage or vice-versa. The registered field patterns are compared with the signals, which have been obtained from
indicate the quality of the lamination insulation and the design the calibration. A loop of a thin wire with the breadth of a few
symmetry of the lamination. millimeters is fixed on the tooth surface and fed with a cali-
bration current (Fig. 7).
D. Evaluation of the Measured Data The difference of the measuring signal is registered with
The recorded values of the tooth leak field can only be and without this loop current and compared with
evalu- ated through the calibration of the measuring coil. The the measured signal of the interlamination short-circuit current
measured (Fig. 8).
84 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 16, NO. 1, MARCH 2001

Also the limit

Fig. 9. Two laminations model: P = Contact load; R = Contact

resistance; R = Ohm resistance of a lamination; u = Voltage
between 2 laminations; = Magnetic flux between 2 laminations.

From that follows the characteristic interlamination short-

circuit magnitude (Product of the interlamination short-
circuit current and the interlamination short-circuit length):

Because for a fixed magnetic induction the voltage between

laminations is constant in the core per unit length, the measuring
signal can then be interpreted as power dissipation from an in-
terlamination short circuit current

This is the maximum possible power dissipation from an

inter- lamination short circuit current. The power on the short
circuit contact depends of the resistance’s in the short-circuit
path.

E. Interlamination Short-Circuit Contact Power

The interlamination short-circuit current of short-circuit cir-
cuited laminations is defined by the induced voltage between
the two laminations and the resistance of the single
laminations. Whether core melting caused by an interlamination
short-circuit arises, depends not only on the interlamination short-
circuit cur- rent, but also on the interlamination short-circuit
resistance, the heat conduction, the heat dissipation and heat
capacity of the contact spot. The maximum interlamination
short-circuit con- tact power between two laminations occurs
when the contact resistance is equal to the lamination
resistance’s 2 . The contact resistance between the back
wedge and the laminations is assumed to be zero. Every
lamination must be well grounded over the back wedge (Fig.
9).
The maximum current between two shorted laminations:

The maximum power dissipation between two shorted lamina-

tions (Fig. 10).
The lamination resistance of large machines is 5–10 m .
The maximum contact current between two 0.5 mm lamina-
tions is about 200 till 400 mA for large turbo-generators (by
the measuring magnetic induction from 0.1 T with a
lamination voltage of about 4 mV. There is evidence that the
limit value of 100 mA from the El Cid test is not correct.
84 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 16, NO. 1, MARCH 2001

Fig. 10. Interlamination short-circuit power curve.

current value of 100 mA can not be used for all machine. The
same interlamination short current value is much less dangerous
for hydromachines than for turbomachines.
The contact power of the interlamination short circuits for
a correct analysis must be know. The maximum contact
power between two 0.5 mm laminations is about 100 mW for
large turbo-generators in operation (magnetic induction 1.5
T with a lamination voltage of about 60 mV). As electrical
machines have similar lamination quality and operating
conditions, the electrical contact power is mainly given by the
yoke height of the stator core. The frequency of the core
melting, which is much higher for turbomachines, confirms
this reflection.

F. The Limit Power Value of the Interlamination Short-Circuit

The interlamination short-circuits between two laminations
with contact thickness of less than 0.1 mm are quickly molten
because of poor heat conduction and heat dissipation to the
surroundings. In the case of short-circuits over many lamina-
tions (higher interlamination short-circuit load), contact melts
can form and trigger core melting. A specification of the limit
values for the maximum permissible interlamination short-
circuit power is difficult because it is hard to predict how the
lamination power will be turned into heat on the surface of
the stator. The local heating of the core depends not only on
the magnitude of the interlamination short-circuit power and
the contact resistance, but also on the geometry of the inter-
lamination short-circuit contact. In the case of more than two
short-circuited laminations, the interlamination short-circuit
power rises approximately as the square of the length and
quickly reaches values which can be considered as dangerous
for the safety of the machine.
Trials and experience show that a maximum
interlamination short-circuit power dissipation in operation of
15 W can rep- resent a danger for machines. An
interlamination short-circuit power dissipation of 15 W is
treated as the lowest level which core melting can arise. This
value should be assumed as the limit value. It is applicable to
all machines.
With this assumed limit value at rated induction, the length
of the short-circuited laminations, which can endanger the
op- erational safety of the machine, would be:
4–10 mm for turbomachines
10–20 mm for
hydromachines.

G. Field Gradient Measurement from an

Interlamination Short-Circuit
A further measurement, with two measuring coils arranged
over each other on the iron stack surface, was often used
to
POSEDEL: INSPECTION OF STATOR CORES IN LARGE MACHINES WITH A LOW YOKE INDUCTION 85
METHOD

— The method shows only hot spots and not all interlami-
nation short-circuits; e.g., an interlamination short-
circuit with a full contact will not have any losses at
the interlamination. Short-circuit spots and therefore the
interlamination short-circuits cannot be heated.
— On generators with assembled stator winding this
method shows only visually accessible hot spots on the
stator sur- face. The other interlamination short-circuits,
which lie on the slot wall or on the slot bottom, cannot
be seen but are even more dangerous.
— To use the temperature as a comparison value for the as-
sessment of the lamination insulation quality is not the
best solution. Less dangerous are interlamination short-
Fig. 11. Iron laminations control with two measuring coils. circuits between two laminations which can have very
high temperatures, in comparison with larger
interlamina- tion short-circuits, which, at small contact
resistance over many laminations show low values.
— The inspection of hot spots is very difficult in large
turbo- generators because of a small-bore diameter and a
large machine length.
— Lamination control during a repair of interlamination
short-circuits requires a very long time. Every interlam-
Fig. 12. Vector diagram of the measured voltages at an interlamination short- ination short-circuit can only be checked after a long
circuit. cooling time of the iron core.

determine the field gradient in addition to the phase and the B. Magnetic Field Measurement of the Interlamination
am- plitude [3] (Fig. 11). Short-Circuit Current with Low Induction
The radial field distribution is strongly dependent on the lo- The low induction methods have several well-recognized
cation of the interlamination short-circuit. At an interlamina- ad- vantages over the high induction method:
tion short-circuit, the fault current over the short-circuited
— The yoke magnetization is possible without large expen-
lam- inations induces in the measuring coils different, phase-
diture and the measuring instrument is simple.
shifted voltages and (Fig. 12).
— The measurement indicates all interlamination short-
The field gradient measurement therefore allows a localiza-
circuits, even those which lie in the slot wall or on the
tion of the interlamination short-circuit. Through comparing the
slot bottom.
radial field gradient of the calibration loop with the field gra-
— The control during a repair of the lamination is easy and
dient of the interlamination short circuit current it is possible
can be carried out immediately.
to determine the fault location. The field gradient of an
— The measurement indicates all interlamination short-
interlam- ination short-circuit on the surface is much more
circuits, even those which are not hot.
pronounced than an interlamination short-circuit in the slots.
— In the case of large turbo-machines, the measurement is
possible with an assembled rotor.
III. COMPARISON OF THE MEASURING METHODS
There are two different measuring systems for the control of
There are two main stator core inspection methods: the stator core insulation with low yoke induction.
— Infrared detection of hot spots on the stator surface at rated — The measuring signal has been interpreted in terms of
yoke induction current [1].
— Magnetic field measurement of the interlamination According to this measurement system, the measuring coil mea-
short- circuit current with low induction. sures directly the interlamination short-circuits current during
the examination of the lamination insulation [1]. The measuring
A. Infrared Detection of Hot Spots on the Stator Surface at coil is considered to be a magnetic potentiometer. But since
Rated Yoke Induction the interlamination short-circuit currents are mostly shorter than
The first, very well known and often used, method at rated the breadth of the measuring coil, it is clear that the direct
yoke induction for recognition of hot spots cannot give measure- ment of the interlamination short-circuit current is
satisfac- tory results, because of the following disadvantages: impossible. Mostly, in comparison with actual interlamination
— In order to carry out the measurements, a high power short-circuit currents, much smaller values are indicated. The
supply source (single phase) is required, as well as measuring coil is, in fact, suitable for the detection of the
exten- sive measuring equipment (long high-current and interlamination short- circuits but not for their quantitative
high- voltage cable, transformer etc.). The method is analysis.
therefore very expensive and time-consuming. — The analysis of the measuring signal with introducing a
calibration procedure.
86 IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 16, NO. 1, MARCH 2001

The magnetic field measurement of the interlamination short- about interlamination short-circuits. The reproducibility of the
circuit current with introducing a calibration procedure is measured values is complete and the measured field pattern
char- acterized by the additional following advantages: serves as a “fingerprint” of the stator core. With the
— A complete qualitative and quantitative assessment of calibration procedure and data processing algorithm,
in- terlamination short-circuits and precise evaluation of assessment criteria (power of the interlamination short-circuit
the measured results. spots) were deter- mined. With the method suggested, the
— The measured phase signal is independent of the yoke quality of inspection should be improved and the operational
in- duction in the range of a magnetic yoke induction safety of the machine increased. This method with a
between meaningful and consistent analysis of the data and assessment
0.003 T and 0.7 T. The measuring signal is fully repro- criteria should be used as a standard control of the stator core
ducible and serves as a “fingerprint” of the stator core lam- by the acceptance tests.
ination state. The measurement is used for trend
analysis. REFERENCES
— The registered field patterns indicate the quality of the
[1] J. Sutton, “El Cid: An easier way to test Stator cores,” Electrical Review,
lamination insulation and the design symmetry of the vol. 207, no. 1, July 4, 1980.
lamination. [2] Z. Posedel, “Method and apparatus for inspecting iron core stacks for
— Interlamination short-circuit power is determined and interlamination shorts,” United States Patent Number 4 996 486, Feb. 26,
1991.
the dangerous spots are indicated. The interlamination [3] G. Biechl, K. Reichert, A. Huber, Z. Posedel, and B. Wiederkehr, “Su-
short- circuit power is the most important value for the pervision of synchronous generators: Condition monitoring and diag-
assess- ment of the danger of an interlamination short- nosis,” in International Conference on the Evolution and Modern As-
pects of Synchronous Machines. Zurich, Switzerland: Swiss Federal
circuit. Institute of Technology, Aug. 27–29, 1991.
— Localization of the fault is possible using the double coil
(measure the radial stray-field gradient.

IV. CONCLUSION
Disadvantages of the classical measuring method of lami- Zlatimir Posedel received the Dipl.Ing. degree in electrical engineering
from the University of Zagreb in 1958. From 1958 to 1963, he was a System
nation insulation control with high yoke induction (high costs, Planning Engineer for Electrical Enterprises Zagreb, and an Assistant
inadequate detection of the interlamination short-circuits, Professor at the University of Zagreb. Since 1963, he has worked with ASEA
unsatisfactory assessment criteria etc.) are avoided by the Brown Boveri, Baden, Switzerland as a Development Engineer. In 1972, he
took part in a special Summer session program on superconducting machinery
measurement of the tooth leak field at low yoke induction. at MIT-Cambridge. He is the Author of several technical papers and holds
With this measurement, all interlamination short-circuits (hot several patents. As co-author he received the IEEE Prize Paper Award 1991
(New York) from the Electric Machinery Committee and the IEEE Paper Award
or not) are detected. The measurement can be carried out 1991 (San Diego) from the Power Engineering Society.
with the assembled rotor. The measurement of the tooth leak
field with the calibration procedure gives correct information